EP3251114A1 - Transcription correction using multi-token structures - Google Patents
Transcription correction using multi-token structuresInfo
- Publication number
- EP3251114A1 EP3251114A1 EP16706043.3A EP16706043A EP3251114A1 EP 3251114 A1 EP3251114 A1 EP 3251114A1 EP 16706043 A EP16706043 A EP 16706043A EP 3251114 A1 EP3251114 A1 EP 3251114A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- hypotheses
- lexicalized
- normalized
- confusion network
- token
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000012937 correction Methods 0.000 title description 8
- 238000013518 transcription Methods 0.000 title description 3
- 230000035897 transcription Effects 0.000 title description 3
- 230000001131 transforming effect Effects 0.000 claims abstract description 12
- 238000010606 normalization Methods 0.000 claims description 38
- 238000003860 storage Methods 0.000 claims description 36
- 238000000034 method Methods 0.000 claims description 34
- 230000008569 process Effects 0.000 claims description 8
- 230000004044 response Effects 0.000 claims description 8
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000002887 multiple sequence alignment Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 description 54
- 230000015654 memory Effects 0.000 description 17
- 238000004891 communication Methods 0.000 description 13
- 238000010586 diagram Methods 0.000 description 5
- 230000000007 visual effect Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 230000002093 peripheral effect Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000013500 data storage Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 241000238558 Eucarida Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000003490 calendering Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000005055 memory storage Effects 0.000 description 1
- 238000003032 molecular docking Methods 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000007723 transport mechanism Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/40—Processing or translation of natural language
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/01—Assessment or evaluation of speech recognition systems
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/083—Recognition networks
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/18—Speech classification or search using natural language modelling
- G10L15/183—Speech classification or search using natural language modelling using context dependencies, e.g. language models
- G10L15/187—Phonemic context, e.g. pronunciation rules, phonotactical constraints or phoneme n-grams
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/28—Constructional details of speech recognition systems
- G10L15/30—Distributed recognition, e.g. in client-server systems, for mobile phones or network applications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/27—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique
- G10L25/33—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique using fuzzy logic
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/18—Speech classification or search using natural language modelling
- G10L15/183—Speech classification or search using natural language modelling using context dependencies, e.g. language models
- G10L15/19—Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules
- G10L15/197—Probabilistic grammars, e.g. word n-grams
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L2015/221—Announcement of recognition results
Definitions
- ASR automated speech recognition
- SLU spoken language understanding
- a challenge in large vocabulary spoken language understanding is robustness to compensate for ASR errors.
- Speech recognition is not perfect, and every user understands that occasional recognition mistakes are a fact of life. From a user's perspective, easiness of correction of recognition mistakes has a substantial impact on an overall experience of a user when speech recognition applications or programs are used. It is with respect to this general technical environment that the present application is directed.
- Examples of the present disclosure describe generation of a multi-arc confusion network to improve, for example, an ability to return alternatives to output generated.
- a confusion network comprising token representations of lexicalized hypotheses and normalized hypotheses is generated.
- Each arc of the confusion network represents a token of a lexicalized hypothesis or a normalized hypothesis.
- the confusion network is transformed into a multi-arc confusion network, wherein the transforming comprising realigning at least one token of the confusion network to span multiple arcs of the confusion network.
- Other examples are also described.
- Figure 1 illustrates an overview of example systems for input recognition and processing.
- Figure 2 illustrates an overview of example determinations by exemplary systems and applications.
- Figure 3 illustrates an example method of generating an example confusion network and an example multi-arc word confusion network
- Figure 4 illustrates an example method of outputting alternative token representations from an example multi-arc confusion network.
- Figure 5 is a block diagram illustrating an example of a computing device with which aspects of the present disclosure may be practiced.
- FIGS. 6A and 6B are simplified block diagrams of a mobile computing device with which aspects of the present disclosure may be practiced.
- Figure 7 is a simplified block diagram of a distributed computing system in which aspects of the present disclosure may be practiced.
- Approaches to correct speech recognition mistakes may use word confusion networks that can offer alternatives for some words once the user indicates that that a hypothesis (e.g., result) provided to the user was not the result the user intended.
- a hypothesis e.g., result
- WCN word confusion network
- Examples of the present disclosure describe generation of a multi-arc token- level confusion network representing hypotheses of recognition results for a spoken utterance to improve an ability to return alternatives to a user for token representations presented to the user.
- the generated multi-arc confusion network comprises token representations resulting from normalization of lexicalized hypotheses. Additionally, the generated multi-arc token-level confusion network is a weighted token-level confusion network with arcs spanning multiples nodes to improve processing in returning alternatives to the user.
- weighted token-level multi-arc confusion networks of the present application are usable to accurately and efficiently return alternatives to the user as it makes obvious alternatives that are available for any selected/identified node span.
- a number of technical advantages are achieved based on the present disclosure including increased efficiency and accuracy in speech recognition processing, increased efficiency and accuracy in processing of alternatives for a user, reduced processing load for follow-up queries from user, increased processing speed, reduction in error rate for speech recognition and improved user interaction performance, among the other plurality of technical advantages.
- FIG. 1 illustrates an overview of example system 100 for input recognition and processing.
- Exemplary system 100 presented is a combination of interdependent components that interact to form an integrated whole for execution of input understanding and processing.
- Components of the systems may be hardware components or software implemented on hardware components of the systems.
- system 100 may include any of hardware components (operating system (OS)), software components (e.g., applications, application programming interfaces, virtual machines) running on hardware components, runtime libraries.
- OS operating system
- software components e.g., applications, application programming interfaces, virtual machines
- runtime libraries e.g., runtime libraries.
- an exemplary system may provide an environment for software components to run, obey constraints set for operating, and makes use of resources or facilities of the system 100, where components may be software (e.g., application, program, module, etc.) running on one or more processing devices.
- an input understanding application or service may be run on a processing device such as a computer, mobile phone, tablet, gaming console, electronic devices, etc., where components of the system may be executed on the processing device.
- components of example systems may be spread across multiple devices.
- input may be entered on a client device (e.g., mobile phone) and information may be processed or accessed from other devices in a network such as a server device and/or data store.
- System 100 also includes components such as data storages/mem ory/stores, for example a knowledge store 108 shown in Figure 1 and described below. Each component may comprise one or more system components. Please refer to Figures 5-7 for additional examples of hardware that may be included as part of the exemplary systems.
- the system 100 includes components such as a speech recognition component 102, a normalization component 104, a confusion network creation component 106 and a knowledge store 108, each having one or more additional components.
- the scale of systems such as system 100 may vary and include more or less components than those described in Figure 1. Interfacing between components of the system 100 occurs through a network connecting components of a system and/or resources external to a system.
- components of system 100 exist locally on a device such as a mobile device.
- components of the system 100 may be connected through a distributed environment such as a cloud computing environment.
- Components of exemplary systems interface with an OS of a processing device to allow input understanding and processing including a user interface (UI) for entering an input and receiving a response.
- system 100 may receive input from a client (e.g., user or service).
- Components of the exemplary system 100 are used to capture and process input and generate and/or update a token confusion network (e.g., to create a multi-arc token confusion network).
- system 100 may operate as an intelligent personal assistant.
- An intelligent personal assistant is a mobile software agent that can perform tasks, or services, for an individual based on user input, location awareness, and the ability to access information from a variety of online sources.
- intelligent personal assistants that the system 100 may operate with include but are not limited to SILVIA, S-Voice, Voice Mate, Google Now, Cortana, Hidi and Siri, among other examples.
- Exemplary system 100 comprises a speech recognition component 102 that is used to capture, process and transcribe a captured input.
- the speech recognition component 102 is able to process spoken queries into text.
- Examples of the speech recognition component 102 include but are not limited to: an automatic speech recognition (ASR), computer speech recognition, or speech to text (STT), among other examples.
- the speech recognition components 102 may interface with an OS of a processing device to provide a user interface for a user to enter input.
- the speech recognition component 102 runs speech recognition on a captured input.
- Speech recognition processing transcribes a received input, for example a speech input, into a lexical recognition result or hypothesis by analyzing an electrical signal (e.g., speech signal) and generating one or more (n-best) results.
- the speech recognition component 102 may utilize the knowledge store 108 when analyzing a captured signal to improve robustness of speech recognition processing to generate the lexical recognition results.
- Lexicalized models are applied to an input signal so that the speech recognition component 102 generates lexicalized results/ hypotheses. For instance, the speech recognition component 102 may transform a captured input signal into a sequence of one or more elements.
- An element is a unit associated with vocabulary such as a letter, symbol, character, word, etc.
- a lexicalized recognition result is "find the closest seven eleven store.”
- the speech recognition component 102 assigns a score or weight to one or more elements of a lexicalized result.
- a score may be associated with an acoustic score for a lexicalized element where the score assesses a probability that the speech signal represents this lexicalized element or with a language model score that assesses a probability that one lexicalized element follows other lexicalized elements.
- a score may be assigned to one or more elements of "find the closest seven eleven store," where the speech recognition component 102 assesses the probability that it accurately interpreted the lexical result.
- each of the results are provided with a separate score derived from the scores of the lexicalized elements it contains.
- the speech component 102 also interfaces directly with the confusion network creation component 106 for output of generated confusion networks and multi-arc confusion networks.
- a confusion network or multi-arc confusion network is generated.
- an input is received where a user requests an alternative for a token of a result returned to a user.
- system 100 processes the request from the user and identifies (e.g., using the speech recognition component 102) that a user would like an alternative representation of one or more tokens identified by the user.
- System 100 outputs alternatives to the user based on evaluation of the multi-arc confusion network.
- a request for alternatives can be provided directly to the confusion network creation component 106 to identify and output alternatives to a system user.
- output of alternative is triggered by a decision of system 100 rather than a request for alternatives from a user.
- Knowledge store 108 is a resource that increases an amount of information that an exemplary system possesses, such as system 100, so that it can make more informed decisions at a given point in time.
- Knowledge store 108 stores any structured and unstructured information that is usable by components of system 100 such as the speech recognition component 102, the
- knowledge data being any data that enhances robustness of an exemplary system or is able to be applied to improve processing by a component of an exemplary system.
- knowledge data stored by knowledge store 108 include but are not limited to: definitions, information for input understanding and transcription, lexical and grapho-phonetical information, rules, name data, historical data (including data from previous queries), interrelationships between components of systems, restrictions (e.g. formal stated descriptions of truth for evaluating data), classes (e.g., collections, concepts, programming types), attribute data (e.g., properties, features, characteristics, parameters), axioms, modeling data including language models, adapted language models,
- the knowledge store 108 may also communicate with components or resources outside of exemplary systems in order to gather or collect information relevant to performing processing and making a best possible decision as to how systems such as system 100 may respond to an input or a request for alternative output.
- the knowledge store 108 may be linked to network resources over the Internet, for example search engines (e.g., Bing, Google Search, Yahoo Search, Ask, WebCrawler, Dogpile, etc.).
- a normalization component 104 interfaces with the speech recognition component 102 to generate normalized results for each lexicalized result generated by the speech recognition component 102.
- the normalization component 104 performs text normalization (e.g., inverse text normalization (ITN)) to a lexicalized result transforming the text of a lexicalized result into a canonical form that a person would write or express (e.g., a readable representation).
- ITN inverse text normalization
- a readable representation of an input of "seven eleven" is "7-11.”
- the normalization component 104 converts each of the n-best recognition alternatives from lexicalized level to token level.
- Each normalized result/hypotheses comprises one or more tokens.
- a token is a unit of communication that is a result of normalization of a lexical element expressing how an output would be presented to a user (e.g., word or alpha-numeric expression).
- the normalization component 104 identifies that a lexicalized result/hypothesis does not require changing. In that case, normalization of a lexicalized result keeps a same lexical representation of a hypothesis as was created by the speech recognition component 102. That is, elements of lexicalized results are the tokens for generation of a confusion network. In the example where a hypothesis/result is "find the closest seven eleven store,” one token may be generated for each element of the lexicalized hypothesis totaling six words. This creates a token representation of a lexicalized hypothesis/result. In another example, normalization processing may generate pseudo-word tokens to include in a token representation. For example, a token of "7-11" is generated as an alternative to tokens representing the words "seven" and "eleven.” In general, each lexicalized result can give rise to one or more alternative normalized results.
- the normalization component 104 is configured to interface with the knowledge store 108 to improve normalization and scoring associated with normalization by accessing resources such as knowledge data of the knowledge store 108.
- the text normalization component 104 provides each normalized result with its own score and combines that score with a score for the lexicalized result from which the normalized result was formed to generate a final score for a normalized result.
- a score generated for one or more tokens created by the normalization is a probability that an alternative token representation is an accurate estimation of a received input.
- normalized results or hypotheses can be assigned new scores based on application of a normalization model (e.g., large scale normalization language model such as BingLM or GoogleLM).
- a robust normalization model is combined with an acoustic score of a lexicalized result to generate a score for one or more tokens of a normalized hypothesis.
- a confusion network creation component 106 interfaces with the normalization component 102 to receive data associated with normalized results for generation of token level confusion networks.
- the confusion network creation component 106 takes recognition results that have been normalized, aligns the recognition results and creates a token level confusion network.
- the confusion network creation component 106 generates a token-level confusion network for the normalized results.
- the confusion network is a sequence of ordered nodes where each pair of consecutive nodes is connected by at least one arc and the sum of probabilities of all arcs leaving one nodes sums up to one (1) or alternatively is associated with a percentage that adds up to one-hundred.
- Each arc of the confusion network is associated with a label (for a token of a normalized result) and a weight (probability/score).
- scores for hypotheses are propagated to individual arcs of the confusion network.
- Knowledge store 108 interfaces with the confusion network creation component 106 to improve robustness for alignment of results and generation of confusion networks where the confusion network creation component 106 has access to resources such as knowledge data of the knowledge store 108.
- the confusion network creation component 106 may align normalized hypotheses/results based on analyzing similarities in parameters associated with one or more tokens of a normalized result.
- alignment of normalized hypotheses/results is executed using dynamic programming.
- exhaustive dynamic programming is used to align the normalized hypotheses/results.
- dynamic programming algorithms used are tailored for practical application of multiple sequence alignment (e.g., exact Multiple Sequence Alignment algorithm, star alignment algorithm, Carrillo Lipman algorithm, etc.).
- the confusion network creation component 106 modifies the created token confusion network, for example, into a multi-arc token confusion network.
- the confusion network creation component 106 re-aligns tokens of generated confusion network to transform a token-level confusion network into a multi-arc token- level confusion network.
- the confusion network creation component 106 re-aligns token representations of a normalized result based on a comparison of time information and grapho-phonetical information associated with a representation (e.g., token or tokens) of a normalized result.
- Grapho-phonetical information used is any information indicating a relationship between written symbols/letters and sounds associated with such symbols/letters.
- grapho-phonetical information examples include but are not limited to spelling (e.g., a spelling for "though” is similar to “tough”), pronunciation (e.g., "tough” is pronounced similar to "puff') and arrangement or sequence of letters in comparison with other tokens, among other examples.
- spelling e.g., a spelling for "though” is similar to “tough”
- pronunciation e.g., "tough” is pronounced similar to "puff'
- arrangement or sequence of letters in comparison with other tokens among other examples.
- the confusion network creation component 106 transforms the generated confusion network into a multi-arc confusion network where nodes remain as aligned before (in the confusion network) but the starting/ending points of individual arcs can be migrated to different nodes searching for the closest one in terms of time information.
- a traditional confusion network encodes possibilities for one-to-one corrections (e.g., an input of "nine” is represented as a token "9”); however many -to-many corrections (e.g., a token "928" as an alternative to individual elements of "9 20 8”) are still non-trivial.
- a confusion network is improved by transforming the confusion network into a multi-arc confusion network.
- Transforming the confusion network into a multi-arc confusion network comprises realigning at least one token of the confusion network to span multiples arcs of the confusion network based on analyzing similarities in time information and grapho-phonetical information of the token associated with an arc of the confusion network.
- the resultant structure offers correction functionality as it makes clearer which alternatives are available for any selected node span. Accordingly, processing of user input is improved as alternative results/hypotheses can be more accurately identified and more efficiently returned to a user, for example, when a user identifies that it would like to receive an alternative to a sequence of several tokens presented to a user.
- Figure 2 illustrates an overview of example determinations by exemplary systems and applications.
- Example systems can be system 100 as described in Figure 1.
- Figure 200 shows an example of results/hypotheses (block 202) output from speech recognition processing (e.g., by speech recognition component 102 as described in Figure 1), results/hypotheses (block 204) output from normalization processing (e.g., by normalization component 104 as described in Figure 1), an example token confusion network (block 206) and an example multi-arc token confusion network (block 208) output from confusion network creation processing (e.g., by confusion network creation component 106 as described in Figure 1).
- speech recognition processing e.g., by speech recognition component 102 as described in Figure 1
- results/hypotheses block 204 output from normalization processing (e.g., by normalization component 104 as described in Figure 1)
- an example token confusion network block 206
- speech recognition processing can yield multiple results or hypotheses for a captured input.
- examples of lexicalized hypotheses determined are shown.
- a score may be associated with elements of the lexicalized hypotheses. Scoring of lexicalized hypotheses/results may be determined based on application of speech recognition language models. As an example, scoring of lexicalized hypotheses may comprise determination of an acoustic score or a language model score as described in the description of Figure 1.
- Normalization processing includes performing ITN or text normalization on the lexicalized hypotheses.
- the normalizing processing comprises generating one or more tokens for elements of each normalized hypothesis.
- normalization processing performed identifies that an element of a lexicalized result/hypothesis does not require changing. In that case, normalization of an element of a lexicalized result keeps a same representation of a hypothesis as was generated by the speech recognition processing by creating one or more tokens are created for elements of the lexicalized hypothesis.
- a normalized hypothesis creates a token representation for lexicalized hypotheses of "Hundred Dollars for New Phone” and “One Hundred Dollars For A New Phone.”
- alternative normalized hypotheses such as "$100 for New Phone” and "$100 for ANew Phone,” are generated as shown in block 204.
- a score may be associated with tokens generated in the normalization processing. For instance, a score may be generated for a normalized result/hypothesis by applying at least one token-level language model to a normalized hypothesis to generate scores for one or more tokens of a normalized hypothesis.
- scoring for a normalized result/hypothesis is generated by rescoring an acoustic score for each of the lexicalized hypotheses to generate a score for each of the normalized hypotheses by applying a token language model to acoustic scores of the lexicalized hypotheses.
- Confusion network creation processing creates a token-level confusion network for the normalized hypotheses as shown in block 206.
- creation of a token-level confusion network comprises aligning normalized hypotheses (or scored normalized results/hypotheses).
- aligning scored responses the aligning performed by the confusion network creation processing is based on evaluation of parameters associated with tokens of normalized hypotheses.
- Arcs (between numbered nodes of block 206) of the generated confusion network have labels (e.g., representing tokens) and scores/weights (shown in parentheses for each arc of block 206, e.g., "(.2)").
- Arcs e.g., "A”, “one”, “ ⁇ eps>” as shown between nodes 1 and 2 of block 206) are labeled with tokens representing alternatives to elements of a hypothesis/result.
- epsilon denoted as “ ⁇ eps>” in block 206) is used to represent a connection between nodes that is null or represents no token.
- Confusion network creation processing further transforms the token-level confusion network into a multi-arc token confusion network as shown in Block 208.
- Block 208 shows a visual representation of an exemplary multi-arc token confusion network.
- Block 208 is a transformation of the token confusion network shown in block 206 and includes an exemplary multi-arc alternative representative of a token (e.g., token of "$100" spans multiple arcs to offer an alternative to "one hundred dollars” or "a hundred dollars”).
- token of "$100" spans multiple arcs to offer an alternative to "one hundred dollars” or "a hundred dollars”
- Figure 3 illustrates an example method 300 of generating an example confusion network and an example multi-arc word confusion network.
- Figure 3 illustrates a method 300 of generating a token-level confusion network and a multi-arc token level confusion network.
- method 300 may be executed by an exemplary system such as system 100 described in Figure 1.
- method 300 may be executed on a device comprising at least one processor configured to store and execute operations, programs or instructions.
- method 300 is not limited to such examples.
- method 300 may be operations performed by an input understanding application or service.
- Flow begins at operation 302, where speech recognition is executed on a captured input.
- Input may be captured and processed by a speech recognition component or module such as the speech recognition component 102 described in Figure 1.
- Lexicalized hypotheses are generated based on processing of the captured input by a speech recognition component or module.
- a generated lexicalized hypothesis comprises one or more elements as described in the description of Figure 1.
- An element is a unit associated with vocabulary such as a letter, symbol, character, word, etc.
- generation of lexicalized hypotheses further comprises generating a score for each of the lexicalized hypotheses.
- an acoustic score and/or a language model score is generated for each of the lexicalized hypotheses.
- a normalized hypothesis (as described in Figure 1) is a result of normalization of a lexicalized hypothesis and comprises one or more tokens as described in the description of Figure 1.
- Normalization processing is performed in operation 304 to generate the normalized hypotheses based on the lexicalized hypotheses or elements of lexicalized hypotheses determined from speech recognition processing.
- token representations are generated for elements of a lexicalized hypothesis/result.
- token representations are newly generated offering alternative representations (e.g., a token of "$1" may be an alternative for the word "dollar") to an element of a lexicalized hypothesis/result.
- Processing of operation 304 may further comprise generating a score for token representations of normalized hypotheses.
- score generation performed in operation 304 comprises generating a final score for each of the normalized hypotheses based on a combination of the score for a lexicalized hypothesis and the score for a normalized hypothesis.
- scoring performed in operation 304 comprises rescoring the acoustic score for each of the lexicalized hypotheses to generate a score for each of the normalized hypotheses by applying a token language model (or adapted token language model) to acoustic scores of the lexicalized hypotheses. Scoring for hypotheses may be generated based on any statistical analysis of the lexicalized and/or normalized hypotheses.
- a confusion network is generated based on the normalization of lexicalized hypotheses.
- each arc of the confusion network generated in operation 306 is a representation of a token of a normalized result.
- Exemplary confusion networks are referenced in Figures 1 and 2 and the accompanying description.
- Generation of the confusion network may comprise aligning tokens of normalized results. For instance, similarities in time information and grapho-phonetical information for tokens are analyzed to initially place arcs of a confusion network.
- Each arc of the generated confusion network is associated with a token and a weight (e.g., score/probability value).
- alignment of hypotheses is completed by dynamic programming approximation that identifies an optimal alignment solution based on evaluation of the time information and the grapho-phonetical information for tokens of the normalized results.
- aligning of tokens is executed using exact Multiple Sequence Alignment algorithms.
- flow proceeds to operation 308 where the confusion network is transformed into a multi-arc confusion network.
- additional processing is performed to transform the confusion network into a multi-arc confusion network.
- the processing performed in operation 308 comprises realigning at least one token of the confusion network to span multiples arcs of the confusion network based on analyzing similarities in time information and grapho-phonetical information of tokens associated with arcs of the confusion network.
- Re-alignment performed in operation 308 comprises changing a starting point and an ending point for an arc associated with the token to span multiple arcs of alternative results/hypotheses.
- Figure 4 illustrates an example method 400 of outputting alternative token representations from an example multi-arc confusion network.
- Figure 4 illustrates a method 400 of using a multi-arc confusion network to provide alternative representations to a user.
- method 400 may be executed by an exemplary system such as system 100 of Figure 1.
- method 400 may be executed on a device comprising at least one processor configured to store and execute operations, programs or instructions.
- method 400 is not limited to such examples.
- method 400 may be performed by an input understanding application or service.
- Flow of method 400 begins at operation 402, where a request is received for alternatives in response to one or more tokens of a token representation presented to a user. For example, when results are returned to a user in response to a spoken utterance, the user may indicate that the output presented is not what the user was looking for or the user may request an alternative to a token presented in the output provided to the user. For instance, an output (e.g., token representation) of "Find the Closest Seven Eleven" may be presented to a user. A user may request alternatives for tokens of "seven" and "eleven.”
- the multi-arc confusion network is used to determine alternatives for one or more tokens.
- the multi-arc confusion network is analyzed to identify an alternative token representation for tokens.
- a token representation of "7-11" may be identified from the multi-arc confusion network as being an alternative to the combination of the words "seven" and "eleven.”
- Flow proceeds to operation 406, where alternative token representations are output from the multi-arc confusion network.
- alternative token representations are output from the multi-arc confusion network.
- an alternative representation identified from the multi-arc confusion network is output to a user.
- a processing device or service using a generated multi-arc confusion network may output one or more
- alternatives to a range of elements or tokens identified by a user of a processing device e.g., mobile device
- service e.g., application or program
- a user may request alternatives to one or more words of a hypothesis/result provided to a user.
- a user may identify that a system or service incorrectly understood a received input and incorrectly interpreted a hypothesis/result.
- Processing for generation or modification of a confusion network may create a new confusion network and/or multi-arc confusion network or alternatively modify an existing confusion network based on additional feedback from a user.
- Figures 5-7 and the associated descriptions provide a discussion of a variety of operating environments in which examples of the invention may be practiced.
- the devices and systems illustrated and discussed with respect to Figures 5-7 are for purposes of example and illustration and are not limiting of a vast number of computing device configurations that may be utilized for practicing examples of the invention, described herein.
- FIG. 5 is a block diagram illustrating physical components of a computing device 502, for example a component of a system with which examples of the present disclosure may be practiced.
- the computing device components described below may be suitable for the computing devices described above.
- the computing device 502 may include at least one processing unit 504 and a system memory 506.
- the system memory 506 may comprise, but is not limited to, volatile storage (e.g., random access memory), non-volatile storage (e.g., read-only memory), flash memory, or any combination of such memories.
- the system memory 506 may include an operating system 507 and one or more program modules 508 suitable for running software applications 520 such as application 528, 10 manager 524, and other utility 526. As examples, system memory 506 may store instructions for execution. Other examples of system memory 506 may components such as a knowledge resource.
- the operating system 507 for example, may be suitable for controlling the operation of the computing device 502. Furthermore, examples of the invention may be practiced in conjunction with a graphics library, other operating systems, or any other application program and is not limited to any particular application or system. This basic configuration is illustrated in Figure 5 by those components within a dashed line 522.
- the computing device 502 may have additional features or functionality.
- the computing device 502 may also include additional data storage devices (removable and/or non-removable) such as, for example, magnetic disks, optical disks, or tape.
- additional storage is illustrated in Figure 5 by a removable storage device 509 and a non-removable storage device 510.
- program modules 508 may perform processes including, but not limited to, one or more of the stages of the operational methods 300-400 illustrated in Figures 3 through 4, for example.
- Other program modules may include electronic mail and contacts applications, word processing applications, spreadsheet applications, database applications, slide presentation applications, input recognition applications, drawing or computer-aided application programs, etc.
- examples of the invention may be practiced in an electrical circuit comprising discrete electronic elements, packaged or integrated electronic chips containing logic gates, a circuit utilizing a microprocessor, or on a single chip containing electronic elements or microprocessors.
- examples of the invention may be practiced via a system-on-a-chip (SOC) where each or many of the components illustrated in FIG. 5 may be integrated onto a single integrated circuit.
- SOC system-on-a-chip
- Such an SOC device may include one or more processing units, graphics units, communications units, system virtualization units and various application functionality all of which are integrated (or "burned") onto the chip substrate as a single integrated circuit.
- the functionality described herein may be operated via application-specific logic integrated with other components of the computing device 502 on the single integrated circuit (chip).
- Examples of the present disclosure may also be practiced using other technologies capable of performing logical operations such as, for example, AND, OR, and NOT, including but not limited to mechanical, optical, fluidic, and quantum
- the computing device 502 may also have one or more input device(s) 512 such as a keyboard, a mouse, a pen, a sound input device, a device for voice input/recognition, a touch input device, etc.
- the output device(s) 514 such as a display, speakers, a printer, etc. may also be included.
- the aforementioned devices are examples and others may be used.
- the computing device 504 may include one or more communication connections 516 allowing communications with other computing devices 518. Examples of suitable communication connections 516 include, but are not limited to, RF transmitter, receiver, and/or transceiver circuitry; universal serial bus (USB), parallel, and/or serial ports.
- Computer readable media may include computer storage media.
- Computer storage media may include volatile and nonvolatile, removable and non- removable media implemented in any method or technology for storage of information, such as computer readable instructions, data structures, or program modules.
- the system memory 506, the removable storage device 509, and the non-removable storage device 510 are all computer storage media examples (i.e., memory storage.)
- Computer storage media may include RAM, ROM, electrically erasable read-only memory (EEPROM), flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other article of manufacture which can be used to store information and which can be accessed by the computing device 502. Any such computer storage media may be part of the computing device 502.
- Computer storage media does not include a carrier wave or other propagated or modulated data signal.
- Communication media may be embodied by computer readable instructions, data structures, program modules, or other data in a modulated data signal, such as a carrier wave or other transport mechanism, and includes any information delivery media.
- modulated data signal may describe a signal that has one or more
- communication media may include wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, radio frequency (RF), infrared, and other wireless media.
- wired media such as a wired network or direct-wired connection
- wireless media such as acoustic, radio frequency (RF), infrared, and other wireless media.
- FIGS. 6A and 6B illustrate a mobile computing device 600, for example, a mobile telephone, a smart phone, a personal data assistant, a tablet personal computer, a laptop computer, and the like, with which examples of the invention may be practiced.
- mobile computing device 600 may be implemented as system 100, components of systems 100 may be configured to execute processing methods as described in Figures 3-4, among other examples.
- FIG. 6A one example of a mobile computing device 600 for implementing the examples is illustrated.
- the mobile computing device 600 is a handheld computer having both input elements and output elements.
- the mobile computing device 600 typically includes a display 605 and one or more input buttons 610 that allow the user to enter information into the mobile computing device 600.
- the display 605 of the mobile computing device 600 may also function as an input device (e.g., a touch screen display). If included, an optional side input element 615 allows further user input.
- the side input element 615 may be a rotary switch, a button, or any other type of manual input element.
- mobile computing device 600 may incorporate more or less input elements.
- the display 605 may not be a touch screen in some examples.
- the mobile computing device 600 is a portable phone system, such as a cellular phone.
- the mobile computing device 600 may also include an optional keypad 635.
- Optional keypad 635 may be a physical keypad or a "soft" keypad generated on the touch screen display.
- the output elements include the display 605 for showing a graphical user interface (GUI), a visual indicator 620 (e.g., a light emitting diode), and/or an audio transducer 625 (e.g., a speaker).
- GUI graphical user interface
- the mobile computing device 600 incorporates a vibration transducer for providing the user with tactile feedback.
- the mobile computing device 600 incorporates input and/or output ports, such as an audio input (e.g., a microphone jack), an audio output (e.g., a headphone jack), and a video output (e.g., a HDMI port) for sending signals to or receiving signals from an external device.
- FIG. 6B is a block diagram illustrating the architecture of one example of a mobile computing device. That is, the mobile computing device 600 can incorporate a system (i.e., an architecture) 602 to implement some examples.
- the system 602 is implemented as a "smart phone" capable of running one or more applications (e.g., browser, e-mail, input processing, calendaring, contact managers, messaging clients, games, and media clients/players).
- the system 602 is integrated as a computing device, such as an integrated personal digital assistant (PDA) and wireless phone.
- PDA personal digital assistant
- One or more application programs 666 may be loaded into the memory 662 and run on or in association with the operating system 664.
- the system 602 also includes a non-volatile storage area 668 within the memory 662.
- the non-volatile storage area 668 may be used to store persistent information that should not be lost if the system 602 is powered down.
- the application programs 666 may use and store information in the nonvolatile storage area 668, such as e-mail or other messages used by an e-mail application, and the like.
- a synchronization application (not shown) also resides on the system 602 and is programmed to interact with a corresponding synchronization application resident on a host computer to keep the information stored in the non-volatile storage area 668 synchronized with corresponding information stored at the host computer.
- other applications may be loaded into the memory 662 and run on the mobile computing device 600, including application 528, IO manager 524, and other utility 526 described herein.
- the system 602 has a power supply 670, which may be implemented as one or more batteries.
- the power supply 670 might further include an external power source, such as an AC adapter or a powered docking cradle that supplements or recharges the batteries.
- the system 602 may include peripheral device port 678 that performs the function of facilitating connectivity between system 602 and one or more peripheral devices. Transmissions to and from the peripheral device port 672 are conducted under control of the operating system 664. In other words, communications received by the peripheral device port 678 may be disseminated to the application programs 666 via the operating system 664, and vice versa.
- the system 602 may also include a radio 672 that performs the function of transmitting and receiving radio frequency communications.
- the radio 672 facilitates wireless connectivity between the system 602 and the "outside world," via a
- communications carrier or service provider Transmissions to and from the radio 672 are conducted under control of the operating system 664. In other words, communications received by the radio 672 may be disseminated to the application programs 666 via the operating system 664, and vice versa.
- the visual indicator 620 may be used to provide visual notifications, and/or an audio interface 674 may be used for producing audible notifications via the audio transducer 625.
- the visual indicator 620 is a light emitting diode (LED) and the audio transducer 625 is a speaker. These devices may be directly coupled to the power supply 670 so that when activated, they remain on for a duration dictated by the notification mechanism even though the processor 660 and other components might shut down for conserving battery power.
- the LED may be programmed to remain on indefinitely until the user takes action to indicate the powered-on status of the device.
- the audio interface 674 is used to provide audible signals to and receive audible signals from the user.
- the audio interface 674 may also be coupled to a microphone to receive audible input, such as to facilitate a telephone conversation.
- the microphone may also serve as an audio sensor to facilitate control of notifications, as will be described below.
- the system 602 may further include a video interface 676 that enables an operation of an on-board camera 630 to record still images, video stream, and the like.
- a mobile computing device 600 implementing the system 602 may have additional features or functionality.
- the mobile computing device 600 may also include additional data storage devices (removable and/or non-removable) such as, magnetic disks, optical disks, or tape.
- additional storage is illustrated in FIG. 6B by the non-volatile storage area 668.
- Data/information generated or captured by the mobile computing device 600 and stored via the system 602 may be stored locally on the mobile computing device 600, as described above, or the data may be stored on any number of storage media that may be accessed by the device via the radio 672 or via a wired connection between the mobile computing device 600 and a separate computing device associated with the mobile computing device 600, for example, a server computer in a distributed computing network, such as the Internet.
- a server computer in a distributed computing network such as the Internet.
- data/information may be accessed via the mobile computing device 600 via the radio 672 or via a distributed computing network.
- data/information may be readily transferred between computing devices for storage and use according to well-known data/information transfer and storage means, including electronic mail and collaborative data/information sharing systems.
- FIG. 7 illustrates one example of the architecture of a system for providing an application that reliably accesses target data on a storage system and handles
- Target data accessed, interacted with, or edited in association with application 528, 10 manager 524, other utility 526, and storage may be stored in different communication channels or other storage types.
- various documents may be stored using a directory service 722, a web portal 724, a mailbox service 726, an instant messaging store 728, or a social networking site 730, application 528, 10 manager 524, other utility 526, and storage systems may use any of these types of systems or the like for enabling data utilization, as described herein.
- a server 720 may provide storage system for use by a client operating on general computing device 502 and mobile device(s) 600 through network 715.
- network 715 may comprise the Internet or any other type of local or wide area network
- client nodes may be implemented as a computing device 502 embodied in a personal computer, a tablet computing device, and/or by a mobile computing device 600 (e.g., a smart phone). Any of these examples of the client computing device 502 or 600 may obtain content from the store 716.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Signal Processing (AREA)
- Fuzzy Systems (AREA)
- Automation & Control Theory (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Machine Translation (AREA)
- Telephonic Communication Services (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US14/606,706 US9384188B1 (en) | 2015-01-27 | 2015-01-27 | Transcription correction using multi-token structures |
PCT/US2016/014411 WO2016122967A1 (en) | 2015-01-27 | 2016-01-22 | Transcription correction using multi-token structures |
Publications (2)
Publication Number | Publication Date |
---|---|
EP3251114A1 true EP3251114A1 (en) | 2017-12-06 |
EP3251114B1 EP3251114B1 (en) | 2018-12-12 |
Family
ID=55410200
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP16706043.3A Active EP3251114B1 (en) | 2015-01-27 | 2016-01-22 | Transcription correction using multi-token structures |
Country Status (4)
Country | Link |
---|---|
US (2) | US9384188B1 (en) |
EP (1) | EP3251114B1 (en) |
CN (2) | CN111508502B (en) |
WO (1) | WO2016122967A1 (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9384188B1 (en) | 2015-01-27 | 2016-07-05 | Microsoft Technology Licensing, Llc | Transcription correction using multi-token structures |
TWI712975B (en) * | 2017-06-29 | 2020-12-11 | 永豐商業銀行股份有限公司 | A voice-activated system for financial information services |
US11043214B1 (en) * | 2018-11-29 | 2021-06-22 | Amazon Technologies, Inc. | Speech recognition using dialog history |
US11817090B1 (en) * | 2019-12-12 | 2023-11-14 | Amazon Technologies, Inc. | Entity resolution using acoustic data |
US11875797B2 (en) * | 2020-07-23 | 2024-01-16 | Pozotron Inc. | Systems and methods for scripted audio production |
US11935519B2 (en) | 2020-10-15 | 2024-03-19 | Google Llc | Preserving speech hypotheses across computing devices and/or dialog sessions |
US11875780B2 (en) * | 2021-02-16 | 2024-01-16 | Vocollect, Inc. | Voice recognition performance constellation graph |
Family Cites Families (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
AU5738296A (en) * | 1995-05-26 | 1996-12-11 | Applied Language Technologies | Method and apparatus for dynamic adaptation of a large vocab ulary speech recognition system and for use of constraints f rom a database in a large vocabulary speech recognition syst em |
US6856956B2 (en) | 2000-07-20 | 2005-02-15 | Microsoft Corporation | Method and apparatus for generating and displaying N-best alternatives in a speech recognition system |
US7571098B1 (en) * | 2003-05-29 | 2009-08-04 | At&T Intellectual Property Ii, L.P. | System and method of spoken language understanding using word confusion networks |
US20070208567A1 (en) | 2006-03-01 | 2007-09-06 | At&T Corp. | Error Correction In Automatic Speech Recognition Transcripts |
JP5366169B2 (en) | 2006-11-30 | 2013-12-11 | 独立行政法人産業技術総合研究所 | Speech recognition system and program for speech recognition system |
CN101647021B (en) * | 2007-04-13 | 2013-03-27 | 麻省理工学院 | Speech data retrieval apparatus, speech data retrieval method, speech data retrieval program and computer usable medium having computer readable speech data retrieval program embodied therein |
US20080319733A1 (en) * | 2007-06-20 | 2008-12-25 | At&T Corp. | System and method to dynamically manipulate and disambiguate confusable speech input using a table |
US8831944B2 (en) * | 2009-12-15 | 2014-09-09 | At&T Intellectual Property I, L.P. | System and method for tightly coupling automatic speech recognition and search |
US8494852B2 (en) | 2010-01-05 | 2013-07-23 | Google Inc. | Word-level correction of speech input |
WO2011149465A1 (en) | 2010-05-27 | 2011-12-01 | Nuance Communications, Inc. | Efficient exploitation of model complementariness by low confidence re-scoring in automatic speech recognition |
JP6066354B2 (en) | 2011-07-01 | 2017-01-25 | 日本電気株式会社 | Method and apparatus for reliability calculation |
KR20130124704A (en) * | 2012-05-07 | 2013-11-15 | 한국전자통신연구원 | Method and apparatus for rescoring in the distributed environment |
CN103514170B (en) * | 2012-06-20 | 2017-03-29 | ***通信集团安徽有限公司 | A kind of file classification method and device of speech recognition |
JP5740368B2 (en) * | 2012-08-28 | 2015-06-24 | 日本電信電話株式会社 | Discriminative speech recognition accuracy estimation apparatus, discriminative speech recognition accuracy estimation method, and program |
CN104157285B (en) * | 2013-05-14 | 2016-01-20 | 腾讯科技(深圳)有限公司 | Audio recognition method, device and electronic equipment |
CN103730115B (en) * | 2013-12-27 | 2016-09-07 | 北京捷成世纪科技股份有限公司 | A kind of method and apparatus detecting keyword in voice |
US9384188B1 (en) | 2015-01-27 | 2016-07-05 | Microsoft Technology Licensing, Llc | Transcription correction using multi-token structures |
-
2015
- 2015-01-27 US US14/606,706 patent/US9384188B1/en active Active
-
2016
- 2016-01-22 CN CN202010307125.4A patent/CN111508502B/en active Active
- 2016-01-22 WO PCT/US2016/014411 patent/WO2016122967A1/en active Application Filing
- 2016-01-22 CN CN201680005243.1A patent/CN107112009B/en active Active
- 2016-01-22 EP EP16706043.3A patent/EP3251114B1/en active Active
- 2016-06-02 US US15/171,149 patent/US9460081B1/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN107112009B (en) | 2020-05-12 |
CN107112009A (en) | 2017-08-29 |
US9460081B1 (en) | 2016-10-04 |
CN111508502A (en) | 2020-08-07 |
US9384188B1 (en) | 2016-07-05 |
CN111508502B (en) | 2023-11-28 |
US20160275071A1 (en) | 2016-09-22 |
WO2016122967A1 (en) | 2016-08-04 |
EP3251114B1 (en) | 2018-12-12 |
US20160217125A1 (en) | 2016-07-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US9460081B1 (en) | Transcription correction using multi-token structures | |
US11688022B2 (en) | Semantic representations using structural ontology for assistant systems | |
JP7317791B2 (en) | Entity linking method, device, apparatus and storage medium | |
US10540965B2 (en) | Semantic re-ranking of NLU results in conversational dialogue applications | |
CN107924483B (en) | Generation and application of generic hypothesis ranking model | |
US10629193B2 (en) | Advancing word-based speech recognition processing | |
US9965465B2 (en) | Distributed server system for language understanding | |
JP6701206B2 (en) | Discriminate ambiguous expressions to improve user experience | |
US11645470B2 (en) | Automated testing of dialog systems | |
CN107111475B (en) | Managing user interactions for input understanding determination | |
US9269354B2 (en) | Semantic re-ranking of NLU results in conversational dialogue applications | |
JP7108675B2 (en) | Semantic matching method, device, electronic device, storage medium and computer program | |
US9171542B2 (en) | Anaphora resolution using linguisitic cues, dialogue context, and general knowledge | |
US10929383B2 (en) | Method and system for improving training data understanding in natural language processing | |
US9361884B2 (en) | Communicating context across different components of multi-modal dialog applications | |
JP2021111334A (en) | Method of human-computer interactive interaction based on retrieval data, device, and electronic apparatus | |
JP7133002B2 (en) | Punctuation prediction method and apparatus | |
US20220129448A1 (en) | Intelligent dialogue method and apparatus, and storage medium | |
WO2016196320A1 (en) | Language modeling for speech recognition leveraging knowledge graph | |
EP4113357A1 (en) | Method and apparatus for recognizing entity, electronic device and storage medium | |
US20220414463A1 (en) | Automated troubleshooter | |
JP2021089728A (en) | Man-machine interaction method, device, and equipment | |
US11170765B2 (en) | Contextual multi-channel speech to text | |
JP7349523B2 (en) | Speech recognition method, speech recognition device, electronic device, storage medium computer program product and computer program | |
KR20240067967A (en) | Voice wake-up method, voice wake-up device, electronic equipment, storage media, and computer program |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
17P | Request for examination filed |
Effective date: 20170524 |
|
AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
AX | Request for extension of the european patent |
Extension state: BA ME |
|
DAV | Request for validation of the european patent (deleted) | ||
DAX | Request for extension of the european patent (deleted) | ||
GRAP | Despatch of communication of intention to grant a patent |
Free format text: ORIGINAL CODE: EPIDOSNIGR1 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: GRANT OF PATENT IS INTENDED |
|
INTG | Intention to grant announced |
Effective date: 20180622 |
|
GRAS | Grant fee paid |
Free format text: ORIGINAL CODE: EPIDOSNIGR3 |
|
GRAA | (expected) grant |
Free format text: ORIGINAL CODE: 0009210 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE PATENT HAS BEEN GRANTED |
|
AK | Designated contracting states |
Kind code of ref document: B1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
REG | Reference to a national code |
Ref country code: GB Ref legal event code: FG4D |
|
REG | Reference to a national code |
Ref country code: CH Ref legal event code: EP |
|
REG | Reference to a national code |
Ref country code: AT Ref legal event code: REF Ref document number: 1077028 Country of ref document: AT Kind code of ref document: T Effective date: 20181215 |
|
REG | Reference to a national code |
Ref country code: DE Ref legal event code: R096 Ref document number: 602016008168 Country of ref document: DE |
|
REG | Reference to a national code |
Ref country code: IE Ref legal event code: FG4D |
|
REG | Reference to a national code |
Ref country code: NL Ref legal event code: FP |
|
REG | Reference to a national code |
Ref country code: LT Ref legal event code: MG4D |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: HR Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: LT Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: BG Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20190312 Ref country code: FI Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: LV Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: ES Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: NO Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20190312 |
|
REG | Reference to a national code |
Ref country code: AT Ref legal event code: MK05 Ref document number: 1077028 Country of ref document: AT Kind code of ref document: T Effective date: 20181212 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: GR Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20190313 Ref country code: RS Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: AL Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: SE Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: PL Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: CZ Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: PT Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20190412 Ref country code: IT Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: SK Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: EE Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: SM Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: RO Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: IS Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20190412 |
|
REG | Reference to a national code |
Ref country code: CH Ref legal event code: PL |
|
REG | Reference to a national code |
Ref country code: DE Ref legal event code: R097 Ref document number: 602016008168 Country of ref document: DE |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: LU Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20190122 |
|
REG | Reference to a national code |
Ref country code: BE Ref legal event code: MM Effective date: 20190131 |
|
PLBE | No opposition filed within time limit |
Free format text: ORIGINAL CODE: 0009261 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT |
|
REG | Reference to a national code |
Ref country code: IE Ref legal event code: MM4A |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: MC Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: SI Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: DK Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 Ref country code: AT Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 |
|
26N | No opposition filed |
Effective date: 20190913 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: BE Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20190131 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: LI Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20190131 Ref country code: CH Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20190131 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: IE Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20190122 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: TR Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: MT Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES Effective date: 20190122 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: CY Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: HU Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT; INVALID AB INITIO Effective date: 20160122 |
|
PG25 | Lapsed in a contracting state [announced via postgrant information from national office to epo] |
Ref country code: MK Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT Effective date: 20181212 |
|
P01 | Opt-out of the competence of the unified patent court (upc) registered |
Effective date: 20230512 |
|
PGFP | Annual fee paid to national office [announced via postgrant information from national office to epo] |
Ref country code: GB Payment date: 20231219 Year of fee payment: 9 |
|
PGFP | Annual fee paid to national office [announced via postgrant information from national office to epo] |
Ref country code: NL Payment date: 20231219 Year of fee payment: 9 Ref country code: FR Payment date: 20231219 Year of fee payment: 9 |
|
PGFP | Annual fee paid to national office [announced via postgrant information from national office to epo] |
Ref country code: DE Payment date: 20231219 Year of fee payment: 9 |